Skip to main content
An aerial view of the Pentagon building in Washington, June 15, 2005. © 2005 Reuters

Eight years ago, it looked like the U.S. military was making progress in reducing the harm its operations caused to civilians. The counterinsurgency strategy in Afghanistan meant winning the hearts and minds of the population. Lessons from the Vietnam War about the strategic imperative to not kill civilians were ubiquitous in conversations with U.S. commanders. With civilian protection seen as critical to a successful mission, the U.S. military was taking steps in Afghanistan and Iraq to investigate civilian casualties, track the aftereffects of its operations, and compensate victims.

“These practices are not perfect, even today, but they represent marked improvements in the conduct of war,” I wrote at the time for Foreign Affairs. As an advocate for minimizing war’s unintended casualties, I had seen these incremental gains accumulate over a decade. The progress wasn’t anywhere near adequate, but it was something. I even went to work as the first (and as yet only) senior adviser on human rights in the office of the chairman of the Joint Chiefs of Staff. Many of us working on this issue felt the United States was turning the corner and could truly become the standard-bearer for protecting civilians in war.

Even as I applauded the U.S. military’s improvement, however, I worried these gains could prove temporary if not properly institutionalized. Painfully learned lessons were being applied on an ad hoc basis, and no one at the Pentagon was “specifically responsible for monitoring civilian harm or figuring out ways to respond to it,” I noted at the time. The U.S. military was at risk of letting its progress slip away.

In 2014, a year after I wrote my essay, a top-secret U.S. strike cell made up of special operations forces was created to go after Islamic State (also known as ISIS) targets in Syria. Over the course of five years, it struck many enemy positions. But a recent New York Times investigation revealed that it also killed an untold number of civilians—deaths that were never officially counted and have only recently been acknowledged by the Defense Department after intense media pressure. Military officials who worked with the task force told the Times that the group would often claim its actions were taken in self-defense to circumvent the rules meant to minimize civilian casualties.

The unearthing of these strikes in Syria came only months after the U.S. military admitted that a high-profile drone strike in Afghanistan last August killed ten civilians, including seven children—and not a suspected terrorist, as the Pentagon first claimed. And in December, the Times released a trove of Defense Department documents that reveal careless targeting, years of civilian deaths, and little accountability in Washington. Together, these events underscore that the U.S. military’s overall record on civilian harm is shameful. Many of these newly revealed strikes appear to be violations of the laws of armed conflict; others represent possible war crimes. Worse yet, many of the problems now coming to light are the same ones that Human Rights Watch, where I work now, and other groups have been documenting for years—to little avail.

In November, U.S. Secretary of Defense Lloyd Austin said the U.S. military “must work harder” to reduce civilian casualties. That pledge has been made before and it’s no longer enough. For anyone to take seriously Austin’s commitment to change, the U.S. military needs to finally address the systemic flaws—problems the Defense Department itself has acknowledged in internal reviews—and allow an independent review of the civilian harm caused by its operations.

I believed the Defense Department back in 2013 when it said protecting civilians was a priority. And when I worked side by side with so many good officers in the Joint Staff, I had hope that their deep belief in how hard the U.S. military tries to avoid killing civilians would make it true. But my trust has been shattered. More important, the credibility of the United States as a moral, lawful actor in armed conflict is in jeopardy.


Civilians will suffer in conflict. That’s a given and should be a major factor for policymakers when considering whether to use force. Even strikes that take every precaution to avoid killing civilians can result in lost lives, lifelong injuries, and crippling damage to infrastructure like homes and water supplies. The U.S. military says it abides by international laws that try to limit these harms, but its actual record shows a pattern of dubious adherence to those laws, repeated mistakes, evidence of recklessness, and outright violations of the law. Even if every single civilian death by U.S. strike were considered legal, the sheer number of them and the similarities between them would be major cause for concern.

The August strike in Kabul was a tragic, recent example of the sorts of flaws that lead to avoidable civilian deaths—flaws well known to anyone who has been paying attention to civilian casualties over the last two decades. In the Kabul case, U.S. drone operators tracked a man named Zemari Ahmadi for eight hours, convinced that he was a terrorist. Days earlier, the ISIS affiliate in Afghanistan, known as ISIS-K, carried out a bomb attack at the Kabul airport, killing an estimated 170 Afghan civilians and 13 U.S. service members. The pressure was on the U.S. military to prevent more attacks. After conducting hours of surveillance, the U.S. military killed Ahmadi and the other civilians around him with a Hellfire missile. But Ahmadi was the wrong man, a humanitarian aid worker going about his ordinary business.

In the wake of the Kabul attack, the Pentagon did what it nearly always does when it is called out in the media for killing civilians. First it denied that civilians were killed, insisting that the strike was “righteous” and the target legitimate—a suspected ISIS suicide bomber. Then, days later, in the face of damning evidence, it called the strike a “mistake” and opened a classified investigation. Finally, as has been the result of nearly every civilian casualty investigation for the last 20 years, the Pentagon found no officials or service members at fault and announced that no one would be held accountable.

Lieutenant General Sami Said, the inspector general of the U.S. Air Force, called the Kabul strike “unique” because of the conditions on the ground in August. It’s true that Kabul was particularly tense after the ISIS-K airport bombing, but there is nothing unique about an edgy environment in theaters of war. The truth is that drone operators did what they have done so many times in the last two decades. They allowed themselves to be misled by what’s known as confirmation bias—the tendency to interpret information in a way that confirms one’s prior beliefs. Fearing another attack by ISIS-K, they mistook water bottles for explosives and Ahmadi for a suicide bomber. The operators were thinking more about the threat and less about the available evidence contradicting the conclusion that Ahmadi was a threat—for example, the fact that he had visited the office of a California-based U.S. humanitarian organization earlier that day.

It would be bad enough if this were an aberration. It isn’t. Larry Lewis, a scientist who led or participated in every official U.S. study on civilian harm, conservatively estimates there’s been at least one civilian casualty incident per week for over 20 years, since the 9/11 attacks in 2001. Nobody knows how many deaths these incidents caused; the U.S. government doesn’t even know. What is known is that many civilians were needlessly killed by a military that publicly carries a banner of moral, legal, and operational superiority into every battle.


Confirmation bias isn’t the only problem plaguing the system. Like many civilians killed by the U.S. military, Ahmadi had been misidentified as a legitimate target. That is, he was not the victim of a strike intended to kill someone else. The positive identification of valid military targets is one of the most critical functions of air strike teams since it leads to decisions about who lives and who dies. Yet such teams have misidentified civilians as legitimate targets in case after case after case. For example, a 2013 report by Lewis described how the majority of civilian casualties caused by U.S. military actions from 2007 to 2012 were the result of misidentification. A 2018 Pentagon study I was part of also contained significant evidence that civilians were misidentified as targets and killed in the 2016 U.S. military campaign to retake Mosul, Iraq, from ISIS and again during the effort to push ISIS out of Raqqa, Syria in 2017.

Another documented flaw in U.S. operations is that drone strikes repeatedly miss the presence of civilian bystanders. Drone operators tend to zoom in when conducting surveillance on a possible target. Although this makes it easier to see the suspect, it also makes it more difficult to see what could be around that target, including civilians coming and going. Strike teams can mitigate this risk by zooming out to look for civilians before launching an attack. The team that carried out the strike in Kabul last August apparently did not do that, judging from a Defense Department fact sheet and comments from officials. Seven children just outside the frame were killed.  

Bad communication during an operation can also lead to civilian deaths. That’s because within a drone operation there is a ground team, a higher command team based elsewhere in the region, an imagery team often based in the United States, and a separate team taking the strike. They’re not necessarily all speaking to each other. So, an imagery analyst trying to discern whether a person is male or female might relay their finding to the command center, but the command center may not share the finding with the trigger-puller or may misinterpret the finding when passing it on. There are ways to make this function better but the U.S. military hasn’t done it. Said, the inspector general who conducted the Kabul strike investigation, indicated that members of the strike cell did not alert others on the team that they were about to launch the attack. The CIA also had eyes on the target in the Kabul strike and saw the children nearby but was not able to get the information to the drone operators in time. The reason why has not been made public.

The most common pattern in U.S. air strikes is how rarely anyone is held responsible for civilian deaths. Repeated internal investigations have resulted in little punishment, perhaps in part because units are investigating themselves. If military personnel and civilian officials knew they might lose their jobs on account of mistakes that led to civilian deaths, they would take greater care to avoid future mistakes. The U.S. military should also reform its justice system to ensure its independence from the chain of command—and criminal liability should be on the table for those who intentionally kill civilians. These and other consequences should be brought to bear in the military coverup of some 70 civilian casualties recently uncovered by a New York Times investigation of a 2019 air strike in Baghuz, Syria, carried out by the classified special operations unit, known as Task Force 9. The Pentagon is conducting an internal investigation of this incident, but it’s magical thinking to expect any accountability.


What’s perhaps most galling about this pattern of civilian death is that U.S. forces have shown progress is possible when there is pressure from the top to improve. Military commanders and civilian defense leaders have grappled with these problems and have found solutions that save lives. When leadership makes protecting civilians a priority, subordinates take it seriously.

In 2005, U.S. commanders in Iraq realized their forces were killing civilians at checkpoints at alarmingly high levels and brought the problem under control by 2006. In 2009, General Stanley McChrystal, the commander of international forces in Afghanistan, made changes to military tactics that led to a 28 percent drop in civilian casualties. Similarly, the average number of civilians killed in errant airstrikes in Afghanistan dropped more than 40 percent after a 2010 tactical directive, a document that provides guidance for troops on the ground, instructed that keeping civilians alive during U.S. military operations was a priority. In 2011, Lewis convinced the U.S. command in Afghanistan to analyze strike data and civilian harm to see if it could lower the risk to civilians. After doing the analysis, he sent the then-commander, General David Petraeus, some suggestions on how to improve, which Petraeus implemented. Civilian deaths started to go back down.

But these changes didn’t last. Over time, the solutions one commander found useful—not just for saving lives but maintaining local support for U.S. military efforts—were eschewed by his replacement. Guidance that once saved civilian lives was loosened to make way for some other priority. As this played out, the military forgot what it learned, or worse, didn’t care enough to make improvements last. For example, in 2009, Admiral Michael Mullen created and led the Joint Staff civilian casualty working group, putting a much-needed spotlight on ways to reduce civilian deaths in Afghanistan. But after some progress, the Pentagon disbanded the group and any improvements were lost.

The same cycle of learning and forgetting happened in the White House, too. Progress made under President Barack Obama—like tightened rules about ensuring civilians weren’t present before a strike—were undone by his successor, President Donald Trump. Indeed, Trump’s overall tone on this issue gave elements of the U.S. military the leeway they wanted to not be so careful about civilian harm. During his campaign for president, Trump promised to “bomb the shit” out of ISIS and target their families, which is a violation of international law. His first defense secretary, retired General James Mattis, told a meeting of U.S. counterterrorism partners that the president directed the U.S. military to “annihilate ISIS.” It’s hard to believe that rhetoric didn’t pave the way to care less about the civilian population in places where the U.S. military was fighting.


Austin’s pledge to “work harder” to limit civilian harm hasn’t produced any tangible steps that give me hope. But the White House is ultimately responsible for setting the tone for how the United States engages in the world and how it fights wars. President Joe Biden has said nothing publicly about the Baghuz or Kabul strikes, although his press office says he supports a full investigation into what went wrong in Kabul. The president requested a review of U.S. counterterrorism policy early in his administration, but he has not yet tasked his defense secretary with fixing the pattern of civilian harm. Biden has not, for example, set up even that most ubiquitous of bureaucratic solutions—an expert working group—to review the past 20 years of civilian harm.

The Pentagon and the White House, over multiple administrations, have shown they cannot—or will not—fix these problems with solutions that stick. There is no other option than to look to Congress for leadership and accountability. But before members of Congress can solve the problems, they need to understand them. They can start by reading the dozens of existing Defense Department studies on the topic. These reports all include recommendations that, had they been implemented, would have spared civilian lives.

Once members of Congress are armed with information, they should hold hearings on the subject. Congress will quickly learn that the Pentagon has no plan in place to curb civilian casualties. Members should ask senior officials for their civilian protection policy (there isn’t one). Ask for their analysis of data on civilian deaths in the past 20 years (that analysis doesn’t exist). Ask what actions have been taken in response to recommendations made over the years on how to prevent civilian harm (very few). Ask who oversees these efforts at the Pentagon and how many resources are devoted to solving the problem of civilian harm (a handful of people are working on civilian harm, none of whom see the issue as their main task, and only pennies on the dollar are dedicated to it, given the Pentagon budget). Ask how many victims’ families have heard directly from the U.S. government (a couple dozen at most in recent years). Ask what the criteria are for holding anyone accountable in cases of mass civilian casualties (there aren’t any). Order them to provide this information and set a prompt deadline. Congress should also remember that these questions only get at civilian casualties caused by the U.S. military and do not cover operations carried out by the CIA.

Congress has failed to ask questions for a long time, but some members are paying attention now. Several committees are considering hearings. Thirty-nine representatives and 11 senators signed a letter to Biden, led by Senators Elizabeth Warren (D-Mass.) and Chris Murphy (D-Conn.), urging him to prioritize civilian protection in new counterterrorism policies. Several more members have privately sent letters to the administration.

In December, John Kirby, the Pentagon’s chief spokesman, blamed the botched August strike in Kabul on "a breakdown in process and execution in procedural events, not the result of negligence, not the result of misconduct, not the result of poor leadership.” This is profoundly and precisely incorrect. The lack of due care is the very definition of negligence, and the lack of leadership and accountability are what allows it to occur again and again.

When The New York Times revealed the truth about who died in the disastrous Kabul strike last August, General Frank McKenzie, the head of U.S. Central Command, remarked that “combat … is an inherently messy, imprecise, bloody business. And we would like for it to be antiseptic. We would like for it to be perfect. It often is just not going to reach those standards of excellence.”

No one is expecting perfection (though, when it comes to deciding who dies, it’s hoped for). But Americans do expect excellence from their armed forces and the civilian leaders whose job it is to oversee them—and there is no reason that standard should not apply to the task of preventing needless civilian deaths.

Your tax deductible gift can help stop human rights violations and save lives around the world.